6006
Biosynthetic Studies with 13C02of Secondary Plant Metabolites. Nicotiana Alkaloids. 1. Initial Experiments C. Richard Hutchinson,* M-T. Stephen Hsia, and Richard A. Carver Contribution from the School of Pharmacy, University of Wisconsin, Madison, Wisconsin 53706. Received October 20, I975
Abstract: The results of an investigation to assess the utility of highly enriched 13C02 as a biosynthetic probe of secondary plant metabolites are described wherein the biosynthesis of nicotine (1) in Nicotiana tabacum and Nicotiana glutinosa is reinvestigated using 97 atom % l3COz coupled with I3C NMR analysis of I3C-enriched 1. It is shown that the incorporation of I3C into 1 as a time function is reflected in the sequence of the appearance of satellite resonances due to lJ,, in the proton noise decoupled 13C NMR spectrum of the alkaloid. Thereby it is possible to identify the portions of 1’s carbon skeleton that arise from different biosynthetic precursors. Unsymmetrical labeling of the N-methylpyrrolidine ring of 1 is observed corroborating earlier findings from I4CO2 based biosynthetic experiments.
The applicability of I3C-labeled compounds to biosynthetic studies has been well documented;’ however, the successful use of I3C-labeled precursors for biosynthetic studies in intact plants has been reported in few instances2 largely because of the prohibitively high precursor dilutions that frequently result therein. The use of I3CO2appeared to us to offer one way of circumventing this dilution problem, as well as the tediousness of label localization by chemical degradation characteristic of I4CO2based experiments. Yet at the outset we did not know if any meaningful biosynthetic information would be obtainable via I3CO2feeding experiments, although in view of results from similar experiments using I4COz3we anticipated that the 13C-labelingpattern of the secondary metabolites would not be uniform and might therefore be reflective of the different units that had been assembled in vivo during the metabolite’s biosynthesis. W e have completed some initial experiments designed to assess the sensibility of using 97 atom % I3CO2as a biosynthetic probe of plant alkaloids and are able to report that quite meaningful results can be obtained when the intramolecular I3C distribution of the natural products is analyzed directly by I3C nuclear magnetic resonance (I3C N M R ) spectroscopy. The biosynthesis of nicotine (1) has been investigated ext e n ~ i v e l yIt. ~is generally believed that 1 is formed in vivo from
fl::
c-c I
I
C’ C.C.*’C,N/C
‘
I
CH,
N” 1
cI
1
C..”C 2
four principal biochemical compounds as depicted by 2. The majority of the experimental results obtained by the use of specifically labeled organic compounds and I4CO2have been interpreted as supporting the biosynthetic pathway adumbrated in Scheme I. Importantly, the equal labeling of C-2’ and C-S’of 1 by [2-14C]ornithine(F) in all cases but one5 and the positive incorporation of putrescine (G) into l6have been accepted as sound evidence that a symmetrical compound (G) is an intermediate of the biosynthetic pathway to 1. In fact, Japanese workers’ have isolated and partially purified enzymes from Nicotiana tabacum that in vitro catalyze the reactions giving G, H , and I from F, although this is not sufficient evidence alone that these enzymes are involved in the biosynthesis of 1 in vivo. Results from biosynthetic studies of 1 obtained using 14C02 wherein the distribution of I4C label in 1 was ascertained by chemical degradation occasionally8 seem to be discordant with those obtained using specifically labeled C and N precursors. Journal of the American Chemical Society
/
98:19
/
Beginning in 1964 Rapoport and co-workers have presented evidence in a set of four major papers9 that in Nicotiana glutinosa label from I4CO2does not always appear in carbons 2’, 3’, 4’,and 5’ of 1 in a manner consistent with a symmetrical biosynthetic intermediate. They did observe that as a function of time (a) the pyridine ring of 1 was more heavily I4C labeled than the N-methylpyrrolidine ring and (b) the specific radioactivity of the N-methyl of 1 usually8 was similar to that of the pyrrolidine ring’s carbons. These experiments have been challenged independently by two groups. Byerrum and co-workersI0 carried out I4CO2 feedings using Nicotiana rustica and N. glutinosa wherein symmetrical I4C labeling of the N-methylpyrrolidine ring of 1 was observed in three separate experiments. They also reported that label from [2-14C]acetate was incorporated into 1loa consistent with the presence of a symmetrical intermediate in the biosynthetic pathway. Leetell showed that one of the chemical degradations used in Rapoport’s work may have been ambiguous and thus could have led to erroneous results in the determination of the I4C-labeling pattern of carbons 4’,5’, and N-CH3 of 1. However, this disclaimer subsequently was refuted by Rapoport12 and in later I4CO2experimentsgea modified degradative sequence of 1 was used by this group apparently to avoid the purported ambiguity in data c01lection.l~One outcome of the work conducted by Rapoport’s group has been the espousal that “the labeling pattern (of 1) via a symmetrical intermediate produced from precursor feedings (other than CO2) might result from a minor or aberrant pathway”.9c Since this particular controversy has been stated to have its genesis in a chemical degradation ambiguity,” we chose to reexamine the biosynthesis of 1 in N. tabacum and N. glutinosa via I3CO2 in which the label distribution would be determined spectroscopically without recourse to degradative techniques. Additionally, such a well-studied, not necessarily well-understood, system offered a good test of the validity and predictability of I3CO2based biosynthetic experiments, which we felt to be necessary if such a technique is to be accepted into the realm of possible ways to study biosynthetic processes in higher plants. It should be obvious that for 13C02to become as useful as I4CO2has been, or to exceed the latter’s applicability in such research, the most difficult hurdle to overcome is the apparent lack of quantifiability of 13CN M R spectral analyses. We address ourselves to this and other questions in the following discussion. Methodology for 13C02 Feedings. All C 0 2 feeding experiments were done in a sealed growth chamber similar to that described by Rapoport et a l l 4 Provision was made for plant removal during feeding experiments by a simple two-door chamber affixed to the front of the growth chamber to minimize loss of labeled COz. Humidity levels inside the growth
September 15, I976
6007 Scheme I. Generally Accepted Biosynthetic Pathway to Nicotine (1)
E
F
chamber rose to saturating levels soon after feeding experiments were commenced and remained there during CO2 feedings. The internal temperature varied between 18 and 23 "C. Illumination levels were from 350 WEinsteins M-2 s-l (top) to 100 PEinsteins M-2 s-l (bottom) of photosynthetically active radiation, which were not saturating for the plants used. In our initial experiments N . tabacum (two experiments) or N. glutinosa (one experiment) plants, 3-5 months old and just prior to their flowering stage, were fed a mixture of 97 atom % 13C02and a tracer quantity of 14C02 for 12-14 h. During the C02 feeding period the atmospheric C 0 2 concentration (0.03%) and specific radioactivity (usually 1.1 X los dpm/mmol) were held constant by continuous replacement of the photosynthetically fixed C02, as monitored by a differential gas flow ir analyzer and a vibrating reed electrometer. At the end of the C 0 2 feeding period the growth lights were turned off and subsequent plant metabolism was maintained in normal atmospheric C 0 2 a t a 14-h day/lO-h night cycle under the same illumination levels. Aliquot groups of plant tops were frozen in liquid N2 upon removal from the growth chamber and chromatographically pure [13C,14C]-1was isolated by standard methods. Its 13C content was measured by E1 mass spectrometry by which the intramolecular I 3C distribution of the N-methylpyrrolidine ring could be accurately measured from the N-methylpyrrolinium fragment ion a t m/e 84,15 but not that of the molecular ion at m/e 163 due to an intense M+ - 1 peak at all voltages tried. Initial determination of the specific radioactivity of [13C,14C]-1 was done to give a rough idea of its I3C enrichment by percentage specific incorporation of 14C;subsequently, after its I3C N M R analysis had been done, accurate determination of its specific radioactivity was done by quantitative dilution with radioinactive 1 and recrystallization of its diperchlorate to constant specific radioactivity. The 13C N M R spectral determinations were done using the Nb-monoethanesulfonate of [13C,14C]-1prepared by titration of the free base in aqueous methanol with 0.05 N aqueous ethanesulfonic acid to pH 5.5 at which only the pyrrolidine nitrogen is protonated.I6 It was found to be absolutely necessary to extract the N M R samples as solutions in D20 with a dilute solution of dithiazone in CC14 to remove paramagnetic metal impurities.17 Otherwise, line broadening and low-signal intensities resulted, particularly for C-2', C-5, and C-3 of 1.
Results and Discussion The technical manipulations associated with the C 0 2 feeding experiments are simple and expedient, once a growth chamber is constructed and the requisite C 0 2 monitoring instruments are obtained. However, this requires more capital expense than is characteristic of 13C-based biosynthetic experiments using bacterial and fungal systems. The size of our
G
I
CH, H I initial growth chamber limited the amount of plants that could be used in C 0 2 feeding experiments, which can be a serious limitation in view of the quantity of pure alkaloid needed for I3C N M R spectral determinations in a reasonable period of time. For the latter reason the first feeding experiments were done using rather large plants, which is not the optimal way to carry out such feedings. As with bacterial and fungal systems it is desirable to add 13C-labeled precursors, whose incorporation is to be determined spectrally, to the system at a time when endogenous precursor and product pools are of minimal size,'* thus reducing the dilution of added precursor, minimizing the resultant time needed to obtain a satisfactory S / N ratio in 13C N M R spectral measurements, and maximizing the sensitivity of the experimental approach. In our case the dilution of the I3C label incorporated into 1 during the course of exposure to I3CO2 and subsequent metabolism in I2C02 was high as reflected by the percentage specific incorporation of 14C02,which increased uniformly from ca. 0.1% after 4 h of the plants' exposure to 13,14C02to 10.2% at 240-h total metabolism time.19 Since the percentage specific incorporation of I4CO2 into 1 represents the statistical average of all nicotine molecules randomly labeled with 14C, a comparable value that can be obtained from the 13CN M R spectrum of 1 is the increase in the total signal intensity per carbon summed over all its ten carbons relative to a natural I3C abundance standard. After examining several derivatives of nicotine we chose to use its Nb-monoethanesulfonate because of the resultant carbon chemical shift dispersion and resolution of the spectra determined in D20. The appropriate spectral data are shown in Table I and typically represented by Figure 1. In order to maximize the reproducibility of the I3C enrichment determined on the basis of total signal intensity enhancement we measured the ratio of the areas of the individual carbon resonances of natural I3C abundance 1 to that of either20 the methyl or methylene of the ethanesulfonate internal reference and compared this ratio to the corresponding one obtained from samples of [13C,14C]-1as their natural I3C abundance ethanesulfonates. This method should minimize inherent differences in the relaxation rates of individual carbons and instrumental error in the precision of the I3C N M R spectral determinations. Rather good precision (footnote e , Table I) among the ratios was obtained in this way, except for the quaternary carbon (C-3) of 1 whose S / N ratio under our spectrometer parameters was very low. The product of these ratios times 1.1% represents the statistical 13Cenrichment a t any carbon in [13C,14C]-1due to the presence in the 13C N M R analysis sample of a random mixture of molecules intermolecularly I3C labeled a t single positions or intramolecularly I3C labeled a t multiple positions, Le., at either C-2,3; (2-43; C-4,5,6; C-2',3'; C-4',5'; or N-CH3. Thus a value of 1.1 in columns 4-6 of Table I represents no 13C enrichment over natural I3C abundance. The percentage total 13C specific incorporation of I3CO2into
Hutchinson, Hsia, Carver / Nicotiana Alkaloids
6008 Table I.
'3C-Labeling Pattern of ['3C,'4C]Nicotine Isolated from I3CO2 Feeding to N . tabacum Relative I3C enrichment. %
Carbon
5 6
150.2 130.7 138.9 126.4 151.5
2'
71
3' 4' 5'
31.5 22.8 51.4 39.5
2 3 4
N-CH3
By Nb-monoethanesulfonate standardC,e
Chemical shift," PPm
By ratio of total area of satellites to singlet aread,e
I Jcc,
Hz
52 hf
122 h
240 h
52 h
122 h
55
1.2
52
g
1.1 g
2.7 2.8 2.9 2.3 2.4 1.8 1.9
h h
h h 47
54 d 54, q 52 58 33 33
1.5 1.5
1.3 1.1 1.1
33
1.3
33
1.1 1.1
1.7 1.5 1.7 1.1
1.2 1.2 1.1
1.1
29 62 29
1.9 2.3
41
i i i i
34Jrk 62-1 57 62 62 65' 58' 49
h
h
64 47
h h h h h
2.2
240 h
Assigned relative to Me& as external standard according to W. 0. Crain, Jr., W. C. Wildman, and J. D. Roberts, J . Am. Chem. Soc., 93,990 (1971), but reversing their assignment for carbons 2 and 6 based on (a) the carbon shift assignments of 1 derived by Birdsall analysis [B. Birdsall, N. J. M. Birdsall, and J. Feeney, J . Chem. Soc., Chem. Commun., 316 (1972)] and (b) the 13Clabeling of these carbons as a function of time. Accurate to 1 3 Hz for the aromatic carbons and f l Hz for the aliphatic carbons. Ratio times 1.1% of the normalized signal area ratios of Nb-mOnOethaneSUlfOnate CH3- or -CH2- of [13C,'4C]-1to corresponding ratios of natural I3C abundance 1. Calculated by eq 4, text. e All peak areas were measured by cutting and weighing of 2X expanded I3C NMR spectra wherein the data precision was C-2 ( i l . 9 ) , C-3 (121.6), C-4 ( & l . l ) , C-5 (&5.3), C-6 (f3.0),C-2'(11.4), C-3'(&4.2),C-4'(&1.9), C-5'(*5.3),andN-CH3 (fl.7),expressed as percent standard deviation of three replicate I3C NMR spectral determinations. f Exposure time to 97 atom % I3CO2(14 h) plus subsequent metabolism in normal CO2. g S/N ratio too low for highly accurate area measurement. No observable satellite signals at comparable S/N ratio. I Doublet satellites observable but signal intensity suspected to be inaccurate. J Assumed no IJ3,4. Assumed no IJ2f.3. Assumed no 1 J 3 , , 4 , . (I
'
nicotine is calculated by summing the values in columns 4-6, Table I, that are greater than 1.1 (the values of C-3 in the 52and 122-h samples were approximated as equal to those of C-2) and then multiplying this sum by 100/97. The specific incorporation of I3C calculated in this way agrees rather well with the specific incorporation of 14C into 1 for the 52-h (1.4 vs. 1.4%) and the 240-h (12.6 vs. 10.2%) samples, but that of the 122-h sample (1.9 vs. 2.7%) does not, which we feel is due to paramagnetic contamination.*' The relative percentage I3C enrichment of the individual carbons of nicotine as a time function should be reflective of the portions of 1 that are derived from biosynthetically distinct precursors. By comparison of these values for the three different samples of [13C,'4C]-1(columns 4-6, Table I), it appears that C-4, C-5, and C-6 become I3Cenriched before the other carbons of 1 in the 52- and 122-h samples, which is consistent with their origin from A. However, in the 240-h sample this distinction is less clear-cut; in fact, now the I3C labeling of carbons 2-6 is of similar magnitude and distinctly larger than that of carbons 2'-5' and the N-CH3. Since it had been reported that label from I4CO2rapidly enters carbons 2,3 and 4,5,6 of 1 uniformly, although not equally between these two units,Ioa we anticipated that each of these carbon sets would be equally I3C enriched in the [13C,14C]-1samples. That this does not appear to be the case from the Nb-monoethanesulfonate derived I3C enrichment data may be a valid observation or may be due to the experimental error inherent in this method. The only independent assessment we can make of the accuracy of this method is to note that the m/e 84 fragment in the mass spectrum of the 240-h sample of [13C,'4C]-1 had a 2.1% I3C enrichment in its M 1 peak, which is similar to the N-CH3 I3C enrichment value (2.3%) calculated from the I3C N M R spectral data. Until we complete a synthesis of [2',3',N-CH3-I3C3]-1, of a known intramolecular I3C distribution, which is presently underway, the question of the accuracy of this method will have to remain unanswered. Presently, we emphasize that this method is at best only indicative of the relative I3C enrichment of the nicotine samples and does not seem to permit distinction quantitatively between singly and multiply labeled molecules present in the alkaloid samples.
+
Journal of the American Chemical Society
Thus, these I3C enirchment values for [13C,14C]-1are not strictly comparable to those derived by the satellite method described below. The most striking feature of the 13CN M R spectrum of the 240-h sample of [13C,'4C]-nicotine(Figure 1 ) is that all resonances, except those of C-5, the N-CH3, and, of course, the internal standard, are flanked by only doublets due to IJcr indicative that a random mixture of essentially doubly I3C labeled molecules is present. This is corroborated by the absence of mass spectral peaks a t m/e 87 and higher mass values for the N-methylpyrrolinium ion of [ 13C,'4C]-1. On the other hand, the C-5 resonance, a doublet and a quartet of ca. equal total intensity, is evidence for a ca. 1:l mixture of molecules labeled a t C-4,5 or C-5,6 and C-4,5,6 present in the sample. Since all of the satellite intensities a r e much greater than expected based on an ca. 2% statistical single carbon 13C enrichment, a mixture of natural 13C abundance 1 and highly I3C-enriched 1 must be present.22 Analogous to the methods thoroughly delineated by London et an equation can be derived that permits calculation of the fractional concentrations of natural abundance and highly 13C-enrichedmaterial present in the 13C N M R analysis sample. (We are indebted to Dr. A. G . McInnes for pointing this out to us.) Thus for two directly bonded carbons, Co-CA I
